The recent advances and unprecedented ubiquity of computation-intensive applications such as virtual reality and mobile augmented reality force new approaches to handle the accompanying challenges. Ultra-Dense Network (UDN) as a leading direction in 5G and beyond offers an opportunistic degree of freedom to be exploited where a massive number of low-power and low-cost Small Cells (SCs) are deployed. In particular, integrating Edge Computing Servers (ECSs) within the SCs at the edge of the cellular network paves the way to tackle several challenges including the processing of computation-intensive tasks with low latency. To exploit the full potentials of UDNs and ECSs, we deploy multiple associations of SCs to the same user while partitioning and offloading its computation-intensive task to the integrated ECSs therein. In this regard, we formulate a mathematical framework using tools from stochastic geometry to evaluate the average processing delay per user. Extensive Monte-Carlo simulations are conducted to verify the accuracy of our analytical results under different system parameters. Results show the existence of an optimal order of multiple-association. In addition, we propose a novel offline task division approach which significantly reduces the overall delay.

中文翻译：

---

超密集网络中的边缘计算和多关联：性能分析


虚拟现实和移动增强现实等计算密集型应用的最新进展和前所未有的普遍性迫使我们采取新的方法来应对随之而来的挑战。超密集网络 (UDN) 作为 5G 及其他领域的主导方向，为部署大量低功耗、低成本小型基站 (SC) 提供了机会主义的自由度。特别是，将边缘计算服务器 (ECS) 集成到蜂窝网络边缘的 SC 中，为解决多项挑战（包括低延迟处理计算密集型任务）铺平了道路。为了充分发挥 UDN 和 ECS 的潜力，我们将多个 SC 关联部署到同一用户，同时将其计算密集型任务分区并卸载到其中的集成 ECS。在这方面，我们使用随机几何工具制定了一个数学框架来评估每个用户的平均处理延迟。进行了广泛的蒙特卡罗模拟，以验证我们在不同系统参数下分析结果的准确性。结果表明存在多重关联的最优顺序。此外，我们提出了一种新颖的离线任务划分方法，可以显着减少总体延迟。